Epilepsy is a common neurological disorder that causes spontaneous recurrent seizures. In spite of major advances in pharmacology, neuroimaging, clinical neurophysiology, and neurosurgery, many patients remain disabled due to uncontrolled seizures. We propose to develop novel diagnostic and therapeutic tools, based on recent discoveries regarding dynamical mechanisms initiating epileptic seizures. We have found characteristic preictal dynamical changes, detectable in the electroencephalogram (EEG), preceding seizures by over 30 minutes (preictal transition, PT). More recently, other investigators have confirmed the presence of PDT. Our research indicates that the PT is demonstrable in the EEG in approximately 90 percent of seizures and that automated paradigms can be used to predict seizures. The potential to predict seizures in advance provides an opportunity to develop innovative diagnostic and therapeutic approaches. Our specific aims are: (1) Specific Aim 1. To continue the development of dynamic measures for the quantification of the spatiotemporal properties of the epileptic transition (years 1-3); (2) To develop specific pattern recognition algorithms for a seizure warning system (SWS) based upon the on-line features of the dynamical properties of brain electrical activity (years 1-4); To implement the dynamic features and pattern recognition algorithms in a SWS for on-line, real-time detection of the preictal dynamical transition (years 2-4); and (4) To evaluate the effects of therapeutic interventions during the preictal transition (years 1-5). The specific spatiotemporal patterns of the PT vary from seizure to seizure and patient to patient. Thus, a sensitive and reliable SWS will require sophisticated signal processing techniques. Dynamical measures will be augmented by other powerful analytic approaches, including multivariate time-series analysis, pattern recognition algorithms, and optimization techniques. To this end, we have gathered experts in signal processing, optimization, V.L.S.I., neurophysiology, neuroanatomy, epilepsy, and neurosurgery. The work will involve the coordination of several research sites throughout the University of Florida Campus including the Brain Dynamics Laboratory (Malcolm Randall V.A. Medical Center), Computer NeuroEngineering Laboratory College of Engineering), Center for Applied Optimization (College of Engineering), an In vitro Neurophysiology Research Laboratory (University of Florida Brain Institute), an In Vivo Neurophysiology Laboratory (Department of Pediatrics) and the Epilepsy Monitoring Laboratory (Shands Hospital). We anticipate that the proposed efforts will result in prototype diagnostic software and devices by the end of year 5. We also will obtain preliminary data that will be used for the design and testing of implantable devices that will activate pulsed therapeutic interventions during the preictal transition.